Multipoint communication techniques used over voice-band phone lines have traditionally relied on a central communications device to manage the use of the line. Communication generally takes place only between the central communications device and all other tributary communications devices. Tributary devices do not communicate directly with each other. The central communications device is connected via a communications channel to every tributary communications device. The central communications device transmits in broadcast fashion to all tributary devices and controls access to the communications channel by one tributary device at a time using polling. There is no contention for transmit access to the shared circuit by the tributaries because access is explicitly granted by the central communications device.
Numerous protocols have been developed to implement this form of polled communications. However, these protocols have the disadvantage of polling overhead. Giving a single tributary device an opportunity to transmit requires polling that device over the communications channel even if the device has nothing to send. The central communications device polls a tributary device by sending a signal to the tributary device to determine if it has data to send. The tributary device then responds to the central communications device by sending a signal over the same communications channel that reveals if the tributary device has data to transmit. This process is performed regardless of whether or not the tributary device has data to send. Consequently, the time used for polling reduces the time available for the transmission of information. Additionally, this polling process has significant overhead in training, timing synchronization, etc. which reduces the total bandwidth or transmission capacity available for the actual information being transmitted.
A number of protocols have been developed for contention access to a common circuit. They are typically intended to allow any communications device to communicate with any other communications device. In these protocols, typically there is no guarantee that only a single device will attempt to transmit at a given time. Therefore, provisions are made to detect simultaneous transmissions and take corrective action when two devices attempt to transmit at the same time over the same channel (also known as a "collision"). In contention access protocols, the frequency of collisions and the overhead associated with collision detection and recovery generally result in significant losses of available bandwidth. Again, the time used to execute these collision detection and recovery techniques reduces the time available for the transmission of information.
Finally, the overhead and complexity of polling can be avoided by using loop protocols operating on specially designed loop circuits. In a loop circuit, the central communications device and all tributary communications devices are connected sequentially in a loop configuration. In this circuit, each device is connected directly to two other devices via a single communications channel that passes through each device. Loop protocols require that the communications channel passes through each communications device so that the device can modify the signal it has received before the signal is transmitted to any other device. Under a loop protocol, communications devices can communicate directly with each other. Usually, data transmitted by one device contains the address of the receiving device. The information is then transmitted to the next device on the loop circuit. This next device looks at the information to see if it is the intended recipient. If it is, the device receives the data. If it is not, the device transmits the data to the next device. This process continues until the intended device receives the transmission. Loop protocols typically circulate a special message called a token. When a station receives the token, it can transmit its own messages (if it has messages to send) in addition to just repeating the transmission received from the previous station in the loop. The token is not transmitted to the next station until the current station is finished transmitting. Loop protocols eliminate the overhead associated with traditional polling and contention access protocols but require the construction of a loop circuit. The added expense of this construction makes a loop circuit less desirable. Loop protocols have the important additional shortcoming that all stations must be given identical access to the shared resource of the circuit bandwidth since the token arrives at each station exactly one time as it makes its way around the loop. It is not practical to allow some stations to use a greater portion of the available bandwidth than others. It may be desirable to allocate more bandwidth to some stations which are supporting applications which require this.